INA301: Overload recovery time specification

Hello Nicolas,

Could you please answer some questions below.

1. What are you using the ALERT pins for? Are you using the output (OUT) pins for the motor control and thus seeing long overload recovery times for the OUT pins of the INA301? Or are you seeing the ALERT pins suffer a delay as well because the OUT pin is also suffering the overload delay?
2. How strong is the overdrive?
3. How fast is the load current changing?

In the application circuit (Figure 42), we are focusing on the ALERT pins and overdriving one of the parts (Part B) to ground with a -10mV sense voltage, while the other part (Part A) is sensing a +10mV. The ALERT pin will respond with normal propagation delay (0.75us to 1.5us according to datasheet) if the internal amplifier's output (OUT pin) is already in the linear region, which is not the case for Figure 42. According to the bidirectional circuit/data (Figure 43) in datasheet, we give ~1.7ms of time for the OUT pin of Part B to leave saturation (Vsense= -1mV, thus OUT driven to GND) and enter the linear range before this Part B has to sense a voltage above its limit.

Thus, the model may not be correctly addressing overload recovery time since your are seeing 10's of milliseconds and according to Figure 43 the overload recovery looks much faster than this; although, the strength of the overdrive signal could be a factor here too.

Best Regards,
Peter Iliya
Current Sensing Applications

Hello Peter,

My intent was to use the INA301 as a short circuit protection (either shoot-through or ground fault protection) for an inverter supplying a brushless motor.

Due to the inverter topology and inverter control scheme, currents through the sense resistor can be either positive or negative in nominal operattion.

Shorts lead to very fast currrent changing slope and a 1 us or less detection time is critical to protect IGBTs from blowing apart.

Simulation seemed to indicate a very slow overload recovery time and my understanding is that you confirm this may take more than1 ms with real parts even with small amount of overdrive voltage.

In such condition, if the nominal current is negative at the time the short occurs and the short leads to a positive current, the detection will occur way too late to protect the IGBTs. If I am right, the INA301 may not be the part to use for my application.

I do not expect the INA303 to give better results as the issue is related to the integrated inamp slow overload recovery.

I may look at implementing the short detection using discrete components instead (high bandwidth opamp + fast comparators).

Please advise if I am wrong with my analysis.

Regards,

NIcolas

Hey Nicolas,

The INA303 actually should solve this problem of internal amplifier output saturation during bidirectional current monitoring. The INA303 has a reference pin that will bias the output stage to that voltage to the reference voltage, most usually this is in the middle of its linear region. So for example you could set a reference voltage of 2.5V and power the part with 5V. Then you could set one limit (VLIMIT1) to 3V and VLIMIT2 to 2V. Thus positive currents will generate an output voltage above 2.5V and negative currents will generate an output voltage below 2.5V.

In this case the only delay you will experience is from the "Total alert propagation delay" or the "Slew-rate-limited tp". The difference is that the first is considered "small-signal" so the testing condition has an input overdrive of 1mV. What this means is that if VLIMIT=2V and gain = 20V/V, then the input signal is jumping from around 99mV to 101mV. This causes OUT signal to go from1.98V to 2.02V and thus trip the comparator.

Slew-rate-limited delays are limited by the slewing of the output of the internal amplifier which has to drive a full-scale jump, 0.5 to 4.5V, Both specifications times begin at the immediate step response of the sense voltage.

Please see the typical bi-directional application circuit in the datasheet for design information.

Also, please let me know if the INA303 cannot meet your requirements, but overall it should inherently give you better performance compared to using two INA301s.

Best regards,

Peter Iliya

Current Sensing Applications